Hi Statalist,
I know my question isn't strictly related to STATA, but I have noticed a lot of brilliant minds on this forum, so I thought you might be able to help. As it involves my master's thesis, any help I can get would obviously mean a lot to me.
I'm currently writing my master's thesis in Finance and International Business with the following problem statement:
"How does firm size affect the leverage of listed US non-financial firms, and why?"
A major argument in the thesis is, that the conventional firm size variable simply act as a proxy for several distinct determinants of leverage (I refer to these as decomposed size effects) - for instance, larger firms are found to be less likely to go bankrupt. I define a total of five decomposed size effects, but unless you deem it relevant for evaluating the research approach that I propose, I won't bore you with the details. :-)
I propose sort of a restricted/unrestricted approach to analysing why firm size is correlated with leverage:
[/INDENT]
I'm looking forward to hearing your thoughts!
Best regards,
Valdemar
I know my question isn't strictly related to STATA, but I have noticed a lot of brilliant minds on this forum, so I thought you might be able to help. As it involves my master's thesis, any help I can get would obviously mean a lot to me.
I'm currently writing my master's thesis in Finance and International Business with the following problem statement:
"How does firm size affect the leverage of listed US non-financial firms, and why?"
A major argument in the thesis is, that the conventional firm size variable simply act as a proxy for several distinct determinants of leverage (I refer to these as decomposed size effects) - for instance, larger firms are found to be less likely to go bankrupt. I define a total of five decomposed size effects, but unless you deem it relevant for evaluating the research approach that I propose, I won't bore you with the details. :-)
I propose sort of a restricted/unrestricted approach to analysing why firm size is correlated with leverage:
- In the restricted model, I regress leverage (debt as a fraction of total capital) on a measure of firm size (the log of book assets) and five control variables that have been deemed relevant in previous literature.
- Then I estimate a number of unrestricted models in each of which I include one new variable that I argue is a good proxy for one of the decomposed size effects that I have identified (e.g., Altman's Z-score as a proxy for the probability of default).
- To test whether a particular decomposed size effect (e.g., the probability of default as proxied by the Altman's Z-score) explains why firm size is correlated with leverage, I would then conduct the following Z-test for equality of coefficients for the firm size variable [proposed in Clogg, C. C., Petkova, E., & Haritou, A. (1995)]:
So my question is, whether this approach to analysing why firm size is correlated with leverage is statistically valid? If not, I would highly appreciate any suggestions for a more appropriate research approach.
and where subscript 1 denotes the restricted model and subscript 2 denotes the unrestricted model. If the coefficient of the firm size variable in the unrestricted model is significantly different from the coefficient in the restricted model, I would conclude that the additional variable for the decomposed size effect (e.g., Altman's Z-score) captures part of the explanatory power of the standard firm size variable. This would then provide support for the hypothesis that firm size is correlated with leverage because larger firms have a lower probability of default.
I'm looking forward to hearing your thoughts!
Best regards,
Valdemar
